Consensus Analysis And Controller Design Of Two Typical Multi-agent Systems

Multi-agent system is a centralized and abstract description of many systems with similar graph structure in the real world.It consists of nodes representing each individ?ual.There are certain interactive relationships among network nodes.The characteris?tics of multi-agent system are various,if the relationships are mutually equivalent,they are undirected networks,otherwise they are directed.In the past decade,global consen?sus has been widely used in engineering,and has been deeply discussed and studied by the scientific community.The key is how to design an available controller to achieve our goal,so that the multi-agent system can achieve global consensus,especially when the complex network has a directed topology.To solve this problem,the traditional controller design methods include adaptive control,pinning control and robust control,while the classical proportional integral differential(PID)controller has been success-fully applied in the nonlinear dynamic system.However,in general complex directed networks,the global consensus of multi-agent system under PI or PD controllers has not been studied thoroughly and comprehensively enough.With the help of the classical Lyapunov global stability theory,this thesis stud-ies the global consensus problem of two typical multi-agent systems with or without time-delay.By setting the real-time average of nodes as the goal of consensus control,appropriate PI controllers and PD controllers are designed respectively,so that the two typical multi-agent systems satisfying special conditions can achieve global consensus under a given controller.Then a simple numerical example for simulation is given to illustrate the main results.The contributions of this thesis are mainly embodied in the following parts:1.In the process of setting the real-time average of node state as the goal of consen-sus control,the property of the product of consensus error matrix and Laplacian matrix is found,which is helpful for exploring the controller design problem in the global con-sensus of the two typical multi-agent systems.2.Aiming at the global consensus problem of multi-agent systems,the controller is designed by using the properties of consensus error matrix and Laplacian matrix,and the sufficient conditions for the global consensus of the two typical multi-agent systems under PID controller are obtained theoretically.3.The numerical simulation model is established to verify the theoretical results.